Open zaoruis opened 1 week ago
Hi, this method does not require training. It uses a pre-trained model from segcontrast and creates a graph from the point cloud, using the point-wise features from the model to define the graph edge weights. To train this pre-trained model, all the instructions are available on the segcontrast repo.
Hello, I'm new to this field. Does your code only have inference? Is there any training code?